Anthropic announced the Anthropic Cyber Mission, including the Critical Infrastructure Defense Program with eleven founding partners and the free OSS Scanner service
Anthropic's three-tier Cyber Verification Program
Anthropic Expands Cyber Verification Program to Three Tiers
The program gives vetted security teams access to capabilities, including authorized offensive work like penetration testing and red-teaming, that are otherwise blocked. The distinction between permanently prohibited uses and unlockable dual-use activities sets a documented boundary for what AI-assisted security work Anthropic will and will not enable.
The full picture
Anthropic launched an expanded Cyber Verification Program on October 6, 2026, merging Project Glasswing and the prior CVP into a single program with three tiers: Defense Access, Red Team Access, and Specialized Access. Verified security professionals gain access to Claude Mythos 5.1, Opus 5.5, and Sonnet 5.5. Through Project Glasswing, partner organizations found at least 129,000 verified software vulnerabilities between April and July 2026, with more than 33,000 rated critical- or high-severity. The expansion is available on the Claude Platform, Google Cloud Vertex AI, and Microsoft Foundry, with Amazon Bedrock access limited to Enterprise Frontier Safeguards-eligible customers.
The three tiers differ in eligibility and safeguard levels. Defense Access is open to individuals and organizations; Red Team Access and Specialized Access are for organizations only. Individuals may only apply for Defense Access and must be on a paid plan. Eligibility factors include the nature of the organization's work, Anthropic's ability to verify identity, the legal and regulatory environment, and risk of diversion or compelled access. Anthropic said it aims to respond to applications within seven business days and places applicants at the highest tier supported by the information provided.
CVP draws a line between two categories of blocked activity. Outright prohibited uses, including ransomware development and mass data exfiltration, remain blocked for all organizations regardless of verification status. High-risk dual-use activities such as vulnerability exploitation and offensive tooling can be unlocked for verified organizations. In email security contexts, CVP enables uses such as building phishing simulation campaigns and testing detection against novel attacks. Data retention is required for misuse monitoring; an upcoming Enterprise Frontier Safeguards solution will eventually allow eligible organizations to store data in their own cloud infrastructure. Organizations on Zero Data Retention agreements are not currently eligible.
How it developed
Anthropic announced on October 6 a restructured Cyber Verification Program with three tiers, Defense Access, Red Team Access, and Specialized Access, consolidating the prior program and Project Glasswing into one.
Verified security professionals gain reduced safeguards on Claude Opus 5.5, Sonnet 5.5, Mythos 5.1, and Fable 5.1 for work such as penetration testing and vulnerability research, while ransomware development and mass data exfiltration remain blocked at all tiers. On CyScenarioBench, the Red Team Access tier produced zero blocks and Claude Opus 5.5 completed 34 of 50 tasks, matching its 67.6% baseline, versus full blocking without CVP access. Program partners have found at least 129,000 verified software vulnerabilities, and some told Anthropic that Claude Mythos models increased their vulnerability-finding rate by months or years.
Ironscales published a glossary entry on the CVP, describing the program's dual-use unlocking framework and eligibility criteria.
Sources
Related
- Continues fromAnthropic's tiered cyber verification program
- Merged inAnthropic's Cyber Mission and OSS Scanner
Want this in your inbox?
I send one email each morning with the stories that moved. If you would rather just read here, that works too.
Subscribe free